Few writers have spent as long on a single question as Eliezer Yudkowsky has spent on this one: what happens if we build a machine cleverer than ourselves before working out how to aim it. An American researcher, largely self-taught, he co-founded the Machine Intelligence Research Institute and has been making the case since the early 2000s, when almost nobody in the field would take the meeting.
Many readers know him instead through the online community he built. The blog sequences that became Rationality: From AI to Zombies and the vast, improbably popular fan novel Harry Potter and the Methods of Rationality both come from him, and both function as tutorials in noticing that you are wrong.
He writes in blunt analogies and refuses the hedging that usually softens technical warnings. You will find him overstated or overdue, and rarely anything in between. The polarising register is deliberate: he has spent twenty years watching carefully qualified language get filed away and forgotten.